We have claimed that emotional intelligence (EI) meets traditional standards for an intelligence . Roberts et al. question whether that claim was warranted. The central issue raised by Roberts et al. concerning our work is whether there are correct answers to questions on tests purporting to measure emotional intelligence as a set of abilities. To address this issue (and others they raise), we briefly restate our view of intelligence, of emotion, and of emotional intelligence. We then present arguments for the reasonableness of measuring emotional intelligence as an ability and summarize recent data suggesting that such measures are, indeed, reliable and that scores based on various scoring criteria are substantially intercorrelated.
 

In 1999, we published, "Emotional Intelligence Meets Traditional Standards for an Intelligence" in the journal Intelligence. In that article, we presented a new scale of emotional intelligence (EI): The Multifactor Emotional Intelligence Scale (MEIS) based on a decade of theoretical and empirical work. We argued, on the basis of the MEIS and findings with it, that EI was a lot like a traditional intelligence: First, it could be measured as an ability for which there were correct answers. Second, the domain of EI was sizable in that we could come up with 12 fairly diverse tasks to measure it, everything from recognizing emotion in faces to understanding how emotions might combine to form other emotional experiences.

Third, after administering the test to 503 adults and 229 adolescents, we found that performance on those 12 diverse tasks was correlated positively across samples. A factor analysis of those tasks indicated that they could be defined by one general factor, and that they also fell into three or four subgroups of skills roughly corresponding to our model of emotional intelligence . Finally, EI ability increased with age, at least across the age ranges that we explored in cross-sectional studies.

Roberts et al. have questioned what kind of an intelligence EI might be, if it is an intelligence at all. One should start with the fact that there is a great deal of agreement between their findings and our own. In their words (ms. p. 54):
 

In fact, the MEIS represented a great step forward for us in relation to our earlier ability scales . It included many tasks that Roberts et al., and we, found intercorrelated well. The overall Cronbach's alpha of the factor-based scale representing the entire test was .96 . And, one of the great advances of the MEIS over our earlier, more limited ability measures, was the first attempt to introduce an "expert" criterion for deciding on a correct answer, although at the time, we only had two such experts. In earlier studies, we had relied on general consensus scoring: basically, the degree to which the individual agreed with the response of the general group was the index of correctness. (We also examined a Target form of scoring, in which, say, a person whose face was being examined regarding the emotions expressed on it, reported how she or he felt). In expert scoring, two of the authors provided their own estimation of the optimal answers to the test.

Still, in reviewing the psychometrics of the MEIS, Roberts et al. find much to be concerned about. We recapitulate some of these empirical concerns here because they form the jumping off point for their theoretical concerns. Again, in their words (ms. pp.54-55):
 

Their findings led them to several conceptual points of importance, raised mostly in their discussion. These are (a) that because there is no agreeable criterion for correct and incorrect answers, EI may well fail to qualify as an intelligence, (b) lower order sensory processes (i.e., perceiving emotion) should score lower on any factor analysis, but it does not (ms. p. 56), and the highest-order factor, managing emotions, has the lowest correlation with the full scale.

Roberts et al. then conclude that, perhaps, at least, there are no objective answers to emotional intelligence tests and that it does not qualify as an intelligence. Perhaps, even, emotional intelligence measures some form of conformity rather than a real intelligence, especially when normative scoring systems based on consensus are used. Essentially, Roberts et al.'s concerns all relate to the general issue of "What sort of an Intelligence is EI?" We address such questions by briefly restating our view of emotional intelligence. Then, we address several questions in turn, including: (a) Are there really correct answers on tests of EI, (b) If, as we argue, consensus scoring is ideal, is emotional intelligence merely a measure of conventionality, and, (c) are the tests reliable, and if not, can they be made to be reliable? We continue to assert that EI is, indeed, a traditional intelligence. Some of our arguments are theoretical, others rely on further findings in studies to which we refer .

The Nature of Intelligence

Symposia on intelligence over the years repeatedly conclude that the first hallmark of intelligence is high-level mental ability such as abstract reasoning . That is, intelligence involves such capacities as seeing relationships such as similarities and differences among objects, being able to reason validly within (and across) content domains, and being able to analyze parts and see their relation to each other and to wholes. Abstract reasoning, although the core aspect, is assisted by several other classes of functions; three such adjunct areas are of chief relevance here: an input, a knowledge base, and meta-strategies . These are enumerated in Table 1.

Abstract reasoning cannot take place without an "input" function. Different intelligences are often defined according to what is input and processed. For example, verbal intelligence pertains to reasoning about language; spatial intelligence pertains to reasoning about the positioning and movement of objects in space. Whatever the area, something must get the information, be it verbal, spatial, or emotional, into the system. Second, abstract reasoning is assisted by a well-organized, related body of knowledge: the knowledge base. This was what Cattell originally referred to as crystallized intelligence ; note that Roberts et al. use the term in a somewhat different sense than we do here . Thirdly, there are meta-cognitions, basically, strategies for operating with an intelligence in the context of broader mental life. A meta-cognition might be that it helps, in analyzing a problem, to write down portions of the problem so that not everything needs to be kept in short-term memory.
 

The Nature of Emotion and Emotional Information

There is considerable diversity of opinion as to what emotion is, of course . A reasonably canonical definition, however, might be that an emotion is an organized mental response to an event, that includes physiological, experiential, and cognitive aspects, among others. Of particular importance to us, is that emotions typically occur in response to relationships . One is angry if blocked from attaining a goal, happy if loved by someone who one loves in return, afraid when threatened, and the like. These relationships may be entirely internal, as when one is afraid of oneself and what one might do, or external, as when one admires another person. If emotions often arise in response to relationships, then emotional information is information about certain forms of these relationships.

One critical aspect of emotional information, to us, is its regularity. Compelling research by Ekman supports Darwin's hypothesis that emotional expression had evolved across species . This strongly implies that emotional information, and the capacity to read it, would show some universals across human beings and even closely related mammalian species. Ekman argued that any apparent differences in human emotional expression from area to area could be attributed to the fact that different cultures taught different display rules about when it was appropriate to express one or another feelings. If any further proof of the regularity of emotional information was required, it could be found in the area of artificial intelligence, where cognitive scientists created expert systems that could understand emotions in rudimentary stories . We deal with the nature of emotional information in greater detail, below, as we discuss the issue of the "correct answer" to an emotional intelligence test item.

The Nature of Emotional Intelligence

Our model of emotional intelligence begins with the idea that emotions contain information about relationships. . When a person's relationship with another person or an object changes, so do their emotions toward that person or object. A person who is viewed as threatening is feared, an object that is expected is liked. Whether these relationships are actual, remembered, or even imagined, they are accompanied by the felt signals of relationship status or change that we call emotions. EI, in turn, refers to an ability to recognize the meanings of emotions and their relationships, and to reason and problem-solve on the basis of them. It further involves employing emotions to enhance cognitive activities .

Our own analysis of emotion-related abilities led us to divide EI into four areas of skills . We call these areas "branches" in reference to the diagrams in which they were first introduced. The four branch model which we now use divides EI into four areas: accuracy at (a) perceiving emotions, (b) using emotions to facilitate thought, (c) understanding emotions, and (d) managing emotions in a way that enhances personal growth and social relations. We view a distinction between the second branch (using emotions) and the other three. Whereas Branches 1, 3, and 4 involve reasoning about emotions, Branch 2 uniquely involves using emotions to enhance reasoning. Finally, we view the four branches as forming a hierarchy, with emotional perception at the "bottom" and "management" at the top. Note, however, that this is a hierarchy in terms of personality, not specifically in terms of underlying affective and cognitive processes. In fact, we view our third level as the most cognitively saturated. We view the top, management level, as involving a balance among many factors: motivational, emotional, and cognitive . This four branch model serves as a basis of our current reviews of the field . With this thumb-nail sketch of our model, let us proceed to Roberts et al.'s concerns.

In Roberts et al.'s estimation, our Emotion Management branch should have its highest correlation with general IQ, whereas our Branch 3 (Understanding) actually does. In fact, however, our model supposes that the third, Understanding, branch should have the highest relation to abstract reasoning. To explain this, consider the outline of our theory in Table 2. There, Understanding is most allied with cognitive processing and abstract reasoning. Management, although the "highest" branch, is considered highest by virtue of its position of creating a bridge or an interface more generally with the more global personality system.

Roberts et al. identified as "the most severe psychometric difficulty," with our work to be, "the lack of convergence between expert- and consensus-scored dimensions." (ms. p. 55). Let us begin in earnest with this issue for two reasons: First, it is primary to whether there can be a correct answer to a test of EI, and hence, whether emotional intelligence is a standard intelligence. Second, it is, by their own description, their most important criticism. As their own concern began with the empirical finding that consensus and expert data correlated only r = .26 (p. 56) in their sample, let us start with the empirical issue. Roberts et al. note that "The discrepancies are sufficiently large that they imply that one or other scoring method should be discarded, in that it is hard to envisage modifications that would bring factors that are correlated at less than 0.50 into alignment (p. 61)."

Empirical Concerns

Roberts et al.'s findings suggest the unlikelihood of any eventual convergence between these scoring methods. At the same time, the MEIS was an experimental measure, a first attempt to operationalize our full model of EI, for which the addition of rudimentary expert scoring was intended as an exploration of the possibility of another criterion of correctness. The expert scoring system of the MEIS was not intended as a final expert criterion. We recognized that two authors answering a long test would be unlikely to create the most optimal version of the expert criterion possible. Indeed, Legree has pointed out that individual experts are typically unreliable. His own research indicated that as experts are aggregated they might be expected to approach the general consensus in this domain . At the time our first paper on the MEIS appeared, we argued that expert scoring and consensus scoring converged to some degree, and that because the general group consensus appeared more reliable, and yielded better test factor structure, it should be employed.

More recently, we have developed a new test of emotional intelligence, the Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT) that attempts to improve on the psychometric qualities of the MEIS. Findings from this new scale are reported in an unpublished manuscript presently under review at this journal . That manuscript, which is currently available from the authors, reports two large-sample psychometric studies. The second study concerns the MSCEIT V2.0, a 141 item ability scale that uses tasks similar to those of the MEIS to measure the four branches of emotional intelligence. Like the MEIS, the MSCEIT can be scored according to a general consensus criterion. That is, if .56 of the sample says that there is a moderate amount of happiness in a face, and a participant agrees, his score is incremented by .56.

Also like the MEIS, we used expert scoring for the MSCEIT. Rather than use two authors as experts, however, we asked 21 members of the International Society of Research in Emotion (ISRE) to answer the MSCEIT questions. We then scored the MSCEIT according to an expert-consensus criterion, based on the proportion of experts from ISRE who answered each item in a particular way. As reported in the manuscript, when over 2,000 participants' scores on the MSCEIT were calculated by general- or expert-consensus scoring, the intercorrelation between the two sets of scores was r = .98. That figure is, obviously, well above the r = .26 figure Roberts et al. actually found with the MEIS, and well above even the r = .50 criterion they could imagine seeing in the future.

Theoretical Concerns

This finding does not, by itself, solve whether there are correct answers to questions assessing EI, but it does greatly simplify the issue to understand that both members of the general population, and emotions experts, converge to something quite close to the same answers as to the emotions in a face, or the meaning of an emotion word. The question can now be refocused on: "What does that consensus mean?" And, is this form of determining a correct answer much different than that used in cognitive intelligence tests?

It is worth noting at the outset, that cognitive IQ tests have items that are "objectively scored" freely intermixed with tests that are scored by a (presumably expert) consensus. An objectively scored test would be something like "digits forward" on the WAIS-III, for which participants hear a series of digits (3..5..1) and simply must repeat them. Other tests, however, require some discussion to arrive at the correct answer. As the WAIS-III manual puts it:
 

The issue then, is not whether experts need to be used, but rather, the nature of emotional versus cognitive information, and the nature of emotional versus cognitive experts. Certainly, there are differences in the two. As we discussed in the section on emotion, we view emotional information as partly consistent across cultures, and as indicating aspects of relationships. But can more be said about the difference between emotional and cognitive information?

Similarities and Differences Between Emotional and Cognitive Information

When thinking about general and expert consensus in the two domains (of emotion and of cognition), there seem to be three issues that come into play: (a) domain of application, (b) general consensus, expert consensus convergence, and (c) systematization of knowledge.

Differences in domain of application. In regard to domain of application, emotional information pertains to the human world: the relations of people (and animals) to one another, and their relation to cultural institutions, ideas, artifacts, and socio-emotional contracts, including rules of behavior. Put another way, emotional information applies primarily to matters of how human beings and their evolutionary ancestors survive and interact with the immediately surrounding world. Human beings have likely come to a general consensus about many emotional meanings. This does not imply that there is only one way to feel, or interpret feelings, but rather, that it helps to know how an individual's reactions compare to how most people would respond emotionally to a situation. Such knowledge helps define the general meaning of emotions in regard to relationships: for example, that happiness arises in harmonious environments, fear in response to threat, and anger to goal blockage, and so on.

Cognitive information, by contrast, describes rules that have areas of application far beyond our immediate living world. Although the child's mathematical world may begin in part with counting fingers and toes, the world of mathematics refers to a sometimes imaginary but often useful numerical world of equations, arithmetical functions, and physically possible and impossible spaces, created both for their beauty and their application. Physical laws such as those describing gravity, acceleration, and the like, apply as much to rocks, stars, and other galaxies, as they do to human beings.

Differences in expertise. This difference in domain of application has implications for the type of expertise in each area. Whereas in the physical, general cognitive realm, objects are studied, in the emotional realm, the object of study is people. This gives rise to the second difference, what we refer to as the issue of (b) general-expert convergence. Given that emotional information is bio-social, i.e., biological and learned, the expert becomes the expert to some extent by studying the group consensus, and becoming, as it were, more accurate about the group consensus than other individuals or small groups. For example, already in the areas of emotions, experts can reliably distinguish sincere or real, from false, tense smiles . It is likely, however, that the general consensus can do this as well. The deciding contribution of expertise is probably more cleanly and expertly to identify and distinguish between the two sorts of smiles, and to elucidate how people do this naturally. That is, the general group will be "messier" in identifying the consensus than experts should be. General-expert convergence occurs in the domain of emotions because experts look for the correct answer by, in part, paying attention to the consensual information of the group.

It is easy to come up with instances, in teaching physics, for example, where the group consensus is simply wrong and the expert opinion is correct. In fact, physicists refer to "lay physics" to represent the sometimes incorrect but consensual notions that people hold about the physical qualities and motions of objects. A well-known example of this is the common misconception (at least among young children) that heavier objects fall faster than lighter ones. Here is a case where expert knowledge plainly trumps lay knowledge, as, since Galileo's famous experiments, it has been known that the two sorts of objects fall at the same rate. The difference here is that the expert in physics, say, conducts experimental research in areas (i.e., the behavior of objects, even invisible, sub-atomic objects) of which the general person has no innate or pragmatic experience.

It is likely possible to come up with parallel instances of emotional expertise as the "lay physics" example. Real expertise does exist in this domain, of course. Take a different example involving systematization. Suppose we ask:

The lay answer is likely to be "a," based on early psychological theories about bullying. More recent evidence, however, has suggested that "b" may be the correct answer, as bullies tend to have high self-esteem . Making such a claim, however, troubles emotions researchers more so, we suspect, than it would physicists. This is part because discoveries in psychology are fraught with contention and have often been reversed as more knowledge is accumulated. This, of course, is also true of physics, after all, in the Ancient Greek period, the expert answer to the problem of what sort of objects fell faster, was that heavier objects fell faster. In addition, however, part of the problem relates to what we would call systematization.

Differences in systematization and institutionalization. The third difference between information (and hence, expertise) in the emotional and the cognitive domain is systematization and institutionalization of knowledge. By systematization, we refer to authoritative dictionaries, manuals, descriptions of operations, and other authoritative texts dealing from a consistent viewpoint with the subject matter. Emotional information certainly has been systematized to some degree, but there is less than universal agreement as to the systematizations. For example, one can trace various enumerations of the meanings of emotions through philosophy, psychology, and now, artificial intelligence programs that decipher emotions in stories. Still, these authoritative manuals have not yet gained widespread cultural currency, are not well known, and are only beginning to be taught in schools. In short, they are not entirely culturally sanctioned. Cultural reasons for this lag are described elsewhere.

There has been a higher level of systematization of certain areas of cognitive information than areas of emotional information. Western schools, colleges, and universities focus on areas such as language, literature, and mathematics, in ways that they simply do not in emotions. In part for that reason, more focus has been placed on creating canonical resource materials in those areas. With those canonical resources, these areas appear more fixed, certain, and objective than does emotional information. Imagine, for a moment, that investigators in the field of emotions were required to establish a high school curriculum for teaching about the emotions and their meaning . Committees would meet, commissions would be established, and, ultimately, an authoritative expert body of knowledge would be produced. Biologists, chemists, and physicists have been doing this for years, with the tacit understanding that the body of expert knowledge will change over the years, but that each iteration represents a further approximation of some, ultimate, truth. When this begins happening in emotions, it is likely that the divergence between experts and the general consensus will be easier to describe and detect.

If emotional information turns out to be the general consensus, and is little different from the expert-consensus, then, have we, as Roberts et al. wonder, created a test of conformity? In their words:
 

Let us begin with the point that conventionality of response is a criticism often leveled at intelligence tests more generally. For example, in reading comprehension, the correct answer is "getting the same point" as everybody else has (or as the established, conventional experts have). Indeed, Guilford referred to most existing intelligence tests as involving convergent thinking. In this sense, our test of EI is also convergent. As with other intelligences, however, it is convergent without being either simply or limited to, conventionality.

Our theory of EI begins with perceiving emotions accurately, as other people conventionally do. We do not see this as much different than saying that literary analysis begins with comprehending the basic content in stories, or that spatial relations begins with being able to name shapes. No one argues that such simple identification of basic linguistic meanings or basic shapes is conformity. EI continues with abstract reasoning about emotions. This includes analyzing linguistic terms relevant to emotion, and analyzing alterations in emotional sequences that are likely to occur (e.g., that delay turns frustration into anger). This is full fledged reasoning, different perhaps, but still arguably on a par conceptually, with, say, understanding the proper order of an argument. If the examples of such emotional reasoning are, perhaps, a bit more limited than that found in linguistic or spatial reasoning, it is probably in large part to the relative paucity of systematization in the emotions area. We expect more such systematization in the future, however, as people continue to come to grips with the importance of the area. Even with such systematization as presently exists, we have been able to come up with enough items to write two entirely different EI tests of hundreds of items, each of which works fairly well (i.e., the MEIS and MSCEIT series).

If we believe the test involves intelligence, then where does creativity enter in? Does it not depend on idiosyncratic reactions? As Roberts et al. asked, should the nonconformist -- a writer or artist who finds a new and original way of expressing an emotion -- be deemed emotionally intelligent? Absolutely, but note that discovering a new way of expressing an emotion in music, for example, doesn't necessarily involve inventing new emotional rules, or idiosyncratic emotional reactions, but rather, may involve arranging musical tones in a new way to elicit old or new emotions and their blends. In fact, it is the very capacity of the writer or artist to portray relations among people and to describe things so as to come up with new blends of feelings or transitions among emotions, that in part marks the creative process. A composer rarely exercises creativity in writing music by playing consistently off key; a writer rarely creates deep, new feelings in a reader by failing to understand how a reader would react emotionally in a similar situation to the character.

Consider Averill and Thompson-Knowles' triad task of emotional creativity. In that task, participants were asked to create a story in which there are relationships that represent particular triads of emotions. In response to the emotions "serene, bewildered, and impulsive," one participant, who was rated quite highly on creativity, wrote:
 

Whether or not you consider that to be great art, it was rated high in novelty, authenticity, and creativity. And yet what makes it work is that its anonymous author did understand the meanings of emotion in a "conventional" sense. What was novel was the creation of a new situation to elicit the feelings.

The MEIS, and our subsequent tests, the MSCEIT RV1.0 and V 2.0 are all reliable tests at the full-scale level (where they all possess split-half reliabilities above r = .90). Our aforementioned manuscript , reports MSCEIT V. 2.0 reliabilities at the branch level ranging from r = .79 to .91 using consensus scoring and r = .77 to .90 using expert scoring. The progression of tests: from MEIS, studied by Roberts et al., to our newer tests the MSCEIT RV 1.1 to MSCEIT V 2.0 also showed a gradual rise in reliability at the lowest-level, reliabilities of individual tasks. Whereas using consensus scoring on the 12 tasks of the MEIS yielded individual task reliabilities ranging from alphas of .49 to .94 , similar scoring for the MSCEIT V2.0 yield individual task alphas from r = .65 to .88 .

Concerns about reliability were first raised about our earliest test, measuring one aspect of EI, by Davies, Stankov, and Roberts . That test provided an empirical demonstration of the possibility of EI rather than a fully operationalized test . Given our primary focus at that time, which was to demonstrate the existence of EI psychometrically, we were not too worried about the admittedly modest, alpha of .63, internal consistency of this rather constrained 1990 measure. To us, if EI existed, it is theoretically important, in part because it challenges the field's view of both emotion and intelligence. Whether or not a test was high in reliability was a secondary concern at that time. After all, almost any test's reliability can be enhanced simply by making it longer (Spearman-Brown). That said, we of course wanted tests to be used by others in the research and applied worlds to possess adequate reliability. We worked hard to ensure that the MEIS and MSCEIT all had full-scale reliabilities over .90, and succeeded.

Roberts and his colleagues were presumably satisfied with the MEIS full-scale reliability of r = .96 that we reported in the 1999 article. They also required, however, that all the subcomponents of the test, even at the level of individual tasks, to be of high reliability as well. It is important to note that this has little to do with whether or not EI exists. Rather, it appears to be, for them, an issue of utility. Their perspective, as it is clarified in their article in this issue, appears to be more applied. They wanted to be assured that when people are tested, the scores those individuals obtain (i.e., are told about) are legitimate and accurate reflections of their ability. This is an obviously legitimate concern on their part. We recognize it to be of growing importance as tests such as the MSCEIT are prepared for general use.

Reliability or Accessibility?

Given that test reliability can be improved simply by lengthening the test, why did we not heed Davies et al.'s criticisms early on and make a test longer than the MEIS? We could have taken the MEIS (or the MSCEIT tests) and added items to them to make them longer and more reliable. Instead, we chose to shorten our tests, and build up reliability at the task level through careful item selection. Investigators have begun to, and will need to, correlate the test with real-life criteria. A short, efficient test that provides one, two, and four reliable scores, like MSCEIT V2.0, can stimulate research better and faster than can a longer, more unwieldy and inefficient test that has optimized reliability for every individual task. Through careful item selection, we have actually improved reliability at the task level (this was accomplished in part by dropping four tasks). To adopt Roberts et al.'s perspective for a moment, it is worth comparing the present MSCEIT V2.0 with the original Wechsler Adult Intelligence Scale (WAIS), used until 1981. That scale was written after roughly 4 decades of intelligence testing, and itself replaced the earlier Wechsler-Bellevue Intelligence Test. The 11 WAIS subscale scores ranged in reliability from r = .60 to .96, not so different from the MSCEIT V2.0.

There are a few issues we are not yet prepared to address in this commentary due to the fact that they are empirical questions that are as-of-yet unresolved. To many people, the most important of these issues would probably concern whether the general and expert consensus are uniform or cultural-bound in Western and non-Western societies. There are several additional issues that will chiefly be of interest to intelligence researchers. For example, why does emotional perception load most highly on factors of general EI? Here, our current work suggests that this finding may not hold with our newer tests and therefore we have not commented on it. Another item chiefly of psychometric interest: is the average interrelation among EI tasks lower than that for general cognitive IQ, and if so why? This is an interesting question for which we have no answer at this time.

In this commentary, we have restated our conceptual approach to intelligence, emotions, and emotional intelligence. Roberts et al. have raised some additional concerns about our theory and measurement of emotional intelligence. Those concerns centered, to a large degree, on the low correlation between two methods of scoring the MEIS (general consensus and expert). In particular, they raised the issue of whether there could be correct answers on a test of emotional intelligence. We are appreciative of the fact that Roberts and colleagues are engaged in this research and have raised these issues. They have prompted us to present new data, and to present a further consideration of the theoretical issues involved in scoring such tests. We have referred to important findings indicating that different scoring methods converge at the r = .98 level rather than the r = .26 level suggested by Roberts et al.'s analyses. In addition, we provided a plausible theoretical explanation of the basis on which correct answers to EI tests can be determined.

In addition, we considered some other concerns of Roberts et al., including test reliability. Although the reliability for the full MEIS is .96, it is lower at the level of the 12, individual tasks. We have argued that the reliability issue raised by Roberts et al. is a limited problem, and one that is addressed by our new scale, the MSCEIT V2.0, which is has been available to researchers for nearly a year.

Implications and Future Research

With such an assessment tool, it is now possible to ask not whether emotional intelligence exists, but whether it is important in various realms of our life. We urge caution in this endeavor, as we would expect emotional intelligence to be an important predictor of significant outcomes, but at levels that are typically found in psychological research. In spite of the claims of popular authors , we do not believe that emotional intelligence will prove to be twice as important as analytic intelligence in predicting "success". What research does exist, however, suggests that emotional intelligence is likely to take its place alongside other important psychological variables as an important predictor of performance at school, home, and work. For example, higher EI may predict psychological mindedness and reduced levels of problem behavior such as drug use and interpersonal violence.

The development and understanding of an intelligence requires a number of years of careful scrutiny and research. The most widely used cognitive scales of intelligence, the Wechsler Intelligence scales, are the product of 60 years of research. Moreover, that research itself was initiated only after 40 years of earlier work on the clinical assessment of intelligence. If the history of the study of intelligence is any guide, there is little question that there is much still to be learned about emotional intelligence, and we look forward to continued work on these critical issues. The first 10 years of emotional intelligence research have been both frustrating and rewarding. Both the pace and quality of research in the area reflect the activities of a healthy field.

 Ackerman, P. L. (1997). Personality, self-concept, interests, and intelligence: Which concept doesn't fit? Journal of Personality, 65, 171-204.
 Averill, J. R., & Thompson-Knowles. (1991). Emotional creativity. In K. T. Strongman (Ed.), International review of studies on emotion (Vol. 1, pp. 269-299). London: Wiley.
 Baumeister, R. F. (1997). Evil: Inside human violence and cruelty. New York: W. H. Freeman & Company.
 Brackett, M. (2001). Emotional intelligence and the life space. Unpublished Master's Thesis, University of New Hampshire, Durham, NH.
 Cattell, R. B. (1943). The measurement of adult intelligence. Psychological Bulletin, 40, 153-193.
 Cohen, J. (Ed.). (1997). Educating minds and hearts:  Social emotional learning and the passage into adolescence. New York: Teachers College Press.
 Cohen, J. (Ed.). (2001). Caring classrooms/intelligent schools: The social emotional education of young children. New York: Teachers College Press.
 Darwin, C. (1872/1965). Expression of the emotions in man and animals. New York: Philosophical Library.
 Davies, M., Stankov, L., & Roberts, R. D. (1998). Emotional intelligence: In search of an elusive construct. Journal of Personality and Social Psychology, 75, 989-1015.
 Dyer, M. G. (1983). The role of affect in narratives. Cognitive Science, 7, 211-242.
 Ekman, P. (1973). Darwin and facial expression: A century of research in review. New York: Academic Press.
 Ekman, P. (1985). Telling lies. New York: Norton.
 Elias, M. J., Zins, J. E., Weissberg, R. P., Frey, K. S., Greenberg, M. T., Haynes, N. M., Kessler, R., Schwab-Stone, M. E., & Shriver, T. P. (1997). Promoting social and emotional learning: Guidelines for educators. Alexandria, VA: Association for Supervision and Curriculum Development.
 Formica, S. (1998). Description of the socio-emotional life space: Life qualties and activities related to emotional intelligence. Unpublished Senior Honors Thesis, University of New Hampshire, Durham, NH.
 Frijda, N. H. (2000). The psychologist's point of view. In M. Lewis & J. M. Haviland-Jones (Eds.), Handbook of emotions (pp. 59-74). New York: Guilford Press.
 Goleman, D. (1995). Emotional intelligence. New York: Bantam Books.
 Goleman, D. (1998). Working with emotional intelligence. New York:: Bantam.
 Lazarus, R. (1991). Emotions and adaptation. New York: Oxford University Press.
 Legree, P. I. (1995). Evidence for an oblique social intelligence factor established with a Likert-based testing procedure. Intelligence, 21, 247-266.
 Matarazzo, J. D. (1972). Wechsler's measurement and appraisal of adult intelligence (5th ed.). New York: Oxford University Press.
 Mayer, J. D. (2001). A field guide for emotional intelligence. In J. Ciarrochi & J. P. Forgas & J. D. Mayer (Eds.), Emotional intelligence and everday life. (pp. 3-24). New York: Psychology Press.
 Mayer, J. D., Caruso, D. R., & Salovey, P. (1999). Emotional intelligence meets traditional standards for an intelligence. Intelligence, 27, 267-298.
 Mayer, J. D., DiPaolo, M. T., & Salovey, P. (1990). Perceiving affective content in ambiguous visual stimuli: A component of emotional intelligence. Journal of Personality Assessment, 54, 772-781.
 Mayer, J. D., & Geher, G. (1996). Emotional intelligence and the identification of emotion. Intelligence, 17, 89-113.
 Mayer, J. D., & Mitchell, D. C. (1998). Intelligence as a subsystem of personality: From Spearman's g to contemporary models of hot processing. In W. Tomic & J. Kingma (Eds.), Advances in cognition and educational practice (Vol. 5, pp. 43-75). Greenwich, CT: JAI Press.
 Mayer, J. D., Perkins, D., Caruso, D. R., & Salovey, P. (2001). Emotional intelligence and giftedness. Roeper Review, 23, 131-137.
 Mayer, J. D., & Salovey, P. (1995). Emotional intelligence and the construction and regulation of feelings. Applied and Preventive Psychology, 4, 197-208.
 Mayer, J. D., & Salovey, P. (1997). What is emotional intelligence? In P. Salovey & D. Sluyter (Eds.), Emotional development and emotional intelligence: Implications for educators (pp. 3-31). New York: Basic Books.
 Mayer, J. D., Salovey, P., & Caruso, D. R. (2000a). Emotional intelligence as zeitgeist, as personality, and as a standard intelligence. In R. Bar-On & J. D. A. Parker (Eds.), Handbook of emotional intelligence (pp. 92-117). New York: Jossey-Bass.
 Mayer, J. D., Salovey, P., & Caruso, D. R. (2000b). Models of emotional intelligence. In R. J. Sternberg (Ed.), Handbook of intelligence (pp. 396-420). Cambridge, England: Cambridge University Press.
 Mayer, J. D., Salovey, P., Caruso, D. R., & Sitaraneos, G. (under review). Emotional intelligence meets traditional standards for an intelligence again: Findings from the MSCEIT.
 Roberts, R. D., Zeidner, M., & Matthews, G. (this issue).
 Rubin, M. M. (1999). Emotional intelligence and its role in mitigating aggression: A correlational study of the relationship between emotional intelligence and aggression in urban adolescents. Unpublished Dissertation, Immaculata College, Immaculata, Pennsylvania.
 Salovey, P., Bedell, B., & Detweiler, J. B. (2000). Current directions in emotional intelligence research. In M. Lewis & Haviland-Jones (Eds.), Handbook of emotions (2nd ed., pp. 504-520). New York: Guildford Press.
 Salovey, P., & Mayer, J. D. (1990.). Emotional intelligence. Imagination, Cognition, and Personality, 9, 185-211.
 Salovey, P., Mayer, J. D., & Caruso, D. R. (in press). The positive psychology of emotional intelligence. In S. J. L. C. R. Snyder (Ed.), The handbook of positive psychology. New York: Oxford University Press.
 Salovey, P., Mayer, J. D., Caruso, D. R., & Lopez, P. N. (in press). Measuring emotional intelligence as a set of abilities with the MSCEIT. In S. J. Lopez & C. R. Snyder (Eds.), Handbook of positive psychology assessment. Washington, DC: American Psychological Association.
 Salovey, P., Woolery, A., & Mayer, J. D. (2001). Emotional intelligence: Conceptualization and measurement. In G. J. O. Fletcher & M. S. Clark (Eds.), Blackwell handbook of social psychology: Interpersonal processes (pp. 279-307). Malden, MA: Blackwell.
 Solomon, R. C. (2000). The philosophy of emotions. In M. Lewis & J. M. Haviland-Jones (Eds.), Handbook of emotions (pp. 3-15). New York: Guilford.
 Sternberg, R. J. (1997). The concept of intelligence and its role in lifelong learning and success. American Psychologist, 52, 1030-1045.
 The Psychological Corporation. (1997). WAIS-III, WMS-III Technical Manual. San Antonio, TX: Author.
 
 
 
 

Table 1: A Summary Overview of Parts of Intelligence (after Mayer & Mitchell, 1998, Table 1).
	Verbal Intelligence	Emotional Intelligence
Adjunct Processing:  Meta-Processing	Knowing that writing something down can help one remember it.	Knowing that helping someone may make oneself feel better.
Abstract Processing	Being able to identify the protagonist of a story and compare the individual to other people.	Being able to analyze an emotion and identify its parts and how they combine.
Adjunct Processing:  The Knowledge Base	Having knowledge (and remembering analyses) of prior instances of stories.	Having knowledge (and remembering analyses) of prior instances of feelings.
Adjunct Processing: Inputting Information	Being able to keep long sentences in memory.	Being able to perceive emotions in faces.